5 ms·

I've been mucking with this stuff again and the SDXL + LCM sampling & LoRA makes 1280x800 images in like 2 second, so about a ~5x speed increase for me (so this

I've been mucking with this stuff again and the SDXL + LCM sampling & LoRA makes 1280x800 images in like 2 second, so about a ~5x speed increase for me (so this would be roughly 2x faster than LCM (??, napkin math)). I've found that the method isn't as good at complex prompts. They claim here this can outperform SDXL 1.0 WRT prompt alignment, but I'm curious what their test methodology is. I searched the paper and I couldn't immediately find how it was evaluated. I think these sorts of subjective measurements are fiendishly hard to quantify given the infinity of possible prompts. Still, exciting stuff always happening here, what a time to be alive.

htrp 3y ago
the use case here is really for segmented inpainting. don't like a part of an image? replace it instantly
- thot_experiment 3y ago
  There are SO many use cases! I maintain we're not even scratching the surface here. You could programmatically reskin video based on crowd participation, you could re-texture VR spaces on the fly. The space of cool shit that you can do with this stuff is growing far faster than we're able to explore it right now.
  - __loam 3y ago
    [flagged]
    - refulgentis 3y ago
      ?
      - __loam 3y ago
        [flagged]
      - thot_experiment 3y ago
        I can't reply to dead comment, but I'll reply to yours RE above: > You can rip off other people's work faster than ever! I have views on IP that mean I reject the ripping off premise on it's face, BUT IF I DID ACCEPT IT Who am I ripping off? What artist is being denied work if i reskin a VR world on the fly with AI? nobody was going to be painting a scene's worth of textures in a few seconds, the AI is enabling new cool things that haven't been seen before, y'all are absolutely out to lunch replying to a comment specifically calling out brand new things that weren't possible before with this garbage.
        __loam 3y ago
        It's so interesting to me that hackernews will flag any comment that dissents to the use of this technology. Supposedly it's irrelevant to the conversation. The value of these models is derived from the training set, not the ml model. Take away the training data and the model does nothing. So who cares if you reskin your vr world on the fly with AI? I'd argue many artists whose work was ingested into these models without consent care very much about that. Many of them have voiced their concerns publicly. So yeah, go ahead and use this bullshit. But you don't get to just ignore the ethics around it. If you want people like me to shut up and let you enjoy your automated slop, use licensed training sets. Until then, this technology is built on exploitation and alienation. Also, you don't just get to dismiss intellectual property because you don't like it. It would be awesome if we lived in some utopia where people didn't need to leverage their skills to eat and pay rent. We don't live in that economy, and I don't think corporate AI is going to get us there unless Microsoft shareholders suddenly become bleeding hearts. Nearly every single dev on this site makes their living off of proprietary code, so it's really rich for y'all to just dismiss the concerns of people whose work is being used for this without permission.
        iinnPP 3y ago
        I have the opposite view of you on the IP front (former 'pirate'), but your critique of HN flagging is spot on. How do you interpret modern web design where everything looks like everything else? Is the next boring SPA also problematic IP usage? These are genuine questions and not meant as an irritant. My view on IP is fairly extreme and I appreciate views that discourage such. Nothing you said about artists needing to eat moved the needle, though I can understand why it would for another person.
- zorgmonkey 3y ago
  It has been a while since I last tried, but I never had very good results when I tried inpainting with SDXL in comparison to the SD1.5 inpainting models.
gliptic 3y ago
They seem to be comparing against SDXL 1.0 at 512x512 which makes no sense to me as SDXL 1.0 is horrible at 512x512. > Using four sampling steps, ADD-XL outperforms its teacher model SDXL-Base at a resolution of 512^2 px

htrp 3y ago

the use case here is really for segmented inpainting. don't like a part of an image? replace it instantly

thot_experiment 3y ago
There are SO many use cases! I maintain we're not even scratching the surface here. You could programmatically reskin video based on crowd participation, you could re-texture VR spaces on the fly. The space of cool shit that you can do with this stuff is growing far faster than we're able to explore it right now.
- __loam 3y ago
  [flagged]
  - refulgentis 3y ago
    ?
    - __loam 3y ago
      [flagged]
    - thot_experiment 3y ago
      I can't reply to dead comment, but I'll reply to yours RE above: > You can rip off other people's work faster than ever! I have views on IP that mean I reject the ripping off premise on it's face, BUT IF I DID ACCEPT IT Who am I ripping off? What artist is being denied work if i reskin a VR world on the fly with AI? nobody was going to be painting a scene's worth of textures in a few seconds, the AI is enabling new cool things that haven't been seen before, y'all are absolutely out to lunch replying to a comment specifically calling out brand new things that weren't possible before with this garbage.
      - __loam 3y ago
        It's so interesting to me that hackernews will flag any comment that dissents to the use of this technology. Supposedly it's irrelevant to the conversation. The value of these models is derived from the training set, not the ml model. Take away the training data and the model does nothing. So who cares if you reskin your vr world on the fly with AI? I'd argue many artists whose work was ingested into these models without consent care very much about that. Many of them have voiced their concerns publicly. So yeah, go ahead and use this bullshit. But you don't get to just ignore the ethics around it. If you want people like me to shut up and let you enjoy your automated slop, use licensed training sets. Until then, this technology is built on exploitation and alienation. Also, you don't just get to dismiss intellectual property because you don't like it. It would be awesome if we lived in some utopia where people didn't need to leverage their skills to eat and pay rent. We don't live in that economy, and I don't think corporate AI is going to get us there unless Microsoft shareholders suddenly become bleeding hearts. Nearly every single dev on this site makes their living off of proprietary code, so it's really rich for y'all to just dismiss the concerns of people whose work is being used for this without permission.
        iinnPP 3y ago
        I have the opposite view of you on the IP front (former 'pirate'), but your critique of HN flagging is spot on. How do you interpret modern web design where everything looks like everything else? Is the next boring SPA also problematic IP usage? These are genuine questions and not meant as an irritant. My view on IP is fairly extreme and I appreciate views that discourage such. Nothing you said about artists needing to eat moved the needle, though I can understand why it would for another person.
zorgmonkey 3y ago
It has been a while since I last tried, but I never had very good results when I tried inpainting with SDXL in comparison to the SD1.5 inpainting models.

gliptic 3y ago

They seem to be comparing against SDXL 1.0 at 512x512 which makes no sense to me as SDXL 1.0 is horrible at 512x512. > Using four sampling steps, ADD-XL outperforms its teacher model SDXL-Base at a resolution of 512^2 px